According to HKEPC, which cites sources from Chinese hardware manufacturers, Chinese customs recently notified logistics firms that NVIDIA’s GeForce RTX 5090 D v2 graphics cards won’t be granted import permits, thus barred from entering the Chinese market. What makes this ban unusual is that it didn’t originate from U.S. export controls; instead, it was initiated by the Chinese government itself, leaving NVIDIA completely unprepared. The RTX 5090 D v2 is a China-exclusive product designed specifically for local consumers and cannot be sold elsewhere. Its development came after U.S. export restrictions forced NVIDIA to discontinue the standard RTX 5090 D (featuring 32GB of GDDR7 memory and a 512-bit memory bus). Subsequently, the company released the D v2 variant with reduced specs—24GB of memory and a 384-bit bus—cutting both memory bandwidth and capacity by 25% to comply with regulations restricting AI-related exports.
Currently, hardware makers have no clear idea about the exact rationale behind the ban. Two main theories circulate in the industry: first, the Chinese government views the D v2 as a downgraded, “insulting” compromise product unworthy of the market; second, certain AI firms have modified the D v2 into a 48GB version tailored for heavy-duty AI workloads, significantly exceeding its official specifications and raising concerns among authorities. Given that the D v2 is exclusive to China and has no alternative sales channels, analysts warn that prolonged restrictions might drive surplus stock into gray markets or illicit smuggling routes. Notably, this ban follows closely after a recent U.S.-China summit, during which Trump visited China alongside NVIDIA CEO Jensen Huang amid hopes of easing tech tensions. Yet the H200 AI chip remains under restriction, and now the deliberate rejection of the D v2 adds yet another unpredictable twist to Sino-American tech rivalry.